Kihwan Kim

++Bringing all the performance improvements we also released a new parakeet model with just 110m parameters that achieves numbers close to SOTA model while also nearly 20x times faster. Model Card: https://lnkd.in/gH5Dscxe

18

The Notebook LM (Experimental) is available in Korea, despite the time difference from the US. Leveraging the power of Gemini 1.5 Pro for personal RAG conversations, this innovative tool creates an engaging experience by integrating personal data into AI interactions. Specifically, it summarizes data that individuals are interested in and makes it easy to ask questions, summarize, and make suggestions based on that data. The impact has been significant, with new users moving beyond simple interactions to engage with AI for in-depth content reviews and thoughtful discussions. We look forward to seeing LLM technology evolve to enable faster, more personalized AI interactions through local inference. #innovation #AI #LLM #personalizedAI

1

How breathtaking it is, we are now standing at the turning point of the industrial revolution: “If you run a factory, your throughput directly correlates to your revenue, the quality of service, the number of people who can use your service. We are now in a world where datacenter throughput utilization is vitally important. It was important in the past, but people don’t measure it. Today, every parameter is measured. When something is a factory, its operations directly correlate to the financial performance of the company,” Jensen Huang (COMPUTEX 2024) Watch the insightful keynote by Jensen Huang at COMPUTEX 2024: [Link to the video](https://lnkd.in/dB9aj2qr) #computex2024

2

🚀 Introducing Google Illuminate: Revolutionizing Academic Learning with AI! 🎧💡 Google Illuminate is an exciting new experimental technology that uses AI to create personalized audio content from academic papers. It adapts the key points of select computer science papers into a conversational format using two AI-generated voices. This innovative tool is designed to help learners grasp complex research, especially for those who prefer auditory learning. 📄🎙️ Currently optimized for computer science publications, Illuminate allows users to listen to these AI-generated conversations through a library of curated content. While the technology is still evolving, user feedback plays a critical role in improving its accuracy. Want to explore cutting-edge research through engaging dialogue? Google Illuminate makes it possible! Tap in, listen, and provide your thoughts on the future of AI-driven learning. 📚✨ Link - illuminate.google.com #AI #MachineLearning #AcademicResearch #GoogleIlluminate #EdTech

4

1) making decisions based on #FirstPrinciples, 2) How Jensen Huang realized that #GPU holds the key to #AI and #AcceleratedComputing - schematic for #software, 3)question your own beliefs and assumptions

3

Transformative new research from Xanadu's quantum algorithms team. Every day, we are getting closer to proving quantum computing's industrial advantage over classical methods. Identifying the right problem + optimizing qubit usage (logical & physical) + reducing the number of gates + decreasing circuit depth + efficient error correction + software tuning = quantum computational advantage. We're getting there. "In this work, we present a quantum algorithm based on product formulas for simulating time evolution under a general vibronic Hamiltonian in real space, capable of handling an arbitrary number of electronic states and vibrational modes. We develop the first trotterization scheme for vibronic Hamiltonians beyond two electronic states and introduce an array of optimization techniques for the exponentiation of each fragment in the product formula, resulting in a remarkably low cost of implementation. To demonstrate practical relevance, we outline a proof-of-principle integration of our algorithm into a materials discovery pipeline for designing more efficient singlet fission–based organic solar cells. Based on commutator bounds, we estimate that a 100 femtosecond evolution using a second-order Trotter product formula of a 4-state model of an anthracene-fullerene interface requires 117 qubits and 1.5×107 Toffoli gates in a reduced dimensionality of 11 modes. In its full dimensionality of 246 modes, it requires 1065 qubits and 2.7 × 109 Toffoli gates." https://lnkd.in/eg9ZewTc #quantumcomputing #FTQC #quantumalgorithms

54

2 Comments

Quest Pro Eye Tracking Accuracy Update Recently came across a new paper by Oleg Komogortsev's lab at Texas State University that reported accuracy of Quest Pro eye tracking (https://lnkd.in/gdxe6rTm). I really like the detailed analysis across population and spatial error percentiles. I think this should be setting industry standard for reporting eye tracking performance. Good news for Quest Pro users: Meta rolled out new eye-tracking model in recent June (v66) update. According to our internal benchmarking, the E95 error is reduced by ~25-30%+ across the population vs numbers reported in the paper (which are based on data collected back in 2023). The bad news are that this is likely the last eye-tracking model update for Quest Pro. But as our CEO alluded to, eye tracking sensors are coming back in future Quest devices (https://lnkd.in/gVPRC-XC). So stay tuned!

171

7 Comments

There are many areas where Nvidia’s future growth trajectories remain significantly and consistently underestimated by the market. This article speaks to a few of those.

5

Watch our CEO Jensen Huang's live keynote at #COMPUTEX2024. Discover how the era of #AI is driving a new industrial revolution across the globe.

6

Learn more about Nemotron-4 340B, a family of open models that developers can use to generate synthetic data for training large language models (LLMs) for commercial applications across healthcare, finance, manufacturing, retail and every other industry. #GenerativeAI #DataCenter

9

Learn how developers are using generative AI, NVIDIA NIM microservices, NVIDIA Omniverse and Universal Scene Description (OpenUSD) to build applications and workflows that enable brand-accurate, targeted and efficient advertising at scale.

3